Originally Posted by
bejammin075
Here was my approach to finding the largest impact side count for Wong Halves. I think there’s a lot of good info in this post, but it may be difficult to digest for some. I don’t know how to write code, so these were my primary tools:
Complete EoR tables.
Formulas from Peter Griffin’s Theory Of Blackjack.
MGP CA.
Excel spreadsheets. Huge ones.
First, I investigated the 789 block side count extensively, then discarded the idea for reasons stated elsewhere. Balanced side counts are the way to go. For the analysis I did, I also decided that balancing single ranks against each other was the best, that way the side count is only used where it helps. Combining ranks in a side count causes increased needs for brain power while also degrading the utility of the side count because the ranks lumped together do not always synergize.
If I was going to develop a side count, it may as well be the best, so I decided to find the most valuable possible side count for Wong Halves, which involved a shit load of math. I got Don’s EoR tables into Excel by scanning the EoR pages, using software to convert to text, then I went tediously number by number through the entire thing to get it 100.00% correct. For my overall analysis, some EoRs were still needed, so I used MGP CA to make a totally complete EoR table.
I setup an Excel table to calculate the PE of every hand matchup based on the main system tags. Then I added on calculations for using balanced side counts to optimize every hand matchup. While investigating a particular balanced side count, I’d vary the side count tag multiplier over a range with fine increments before combining those tags with the main system tags, and calculate the PE for the entire range of tag multipliers. For each play I could see the best multiplier of the side count tags that gives the biggest improvement in PE. Then after seeing the optimal multipliers, I decided which integer or half integer multiples would be practical, because this has to lead to a workable strategy that can be used with mental math.
I came up with a very good but approximate formula to rank the impact of each balanced side count, then I systematically went through every possible balanced side count of 2 ranks, determined their impact, and selected the one with the largest impact for further strategy development. I didn’t have the means to do a perfect analysis, but I’m pretty sure a perfect analysis would arrive at the same results in the end. I’m providing a blue print that others can do probably even better than me, and for their particular counting system.
Formula for impact of a balanced side count:
Sum of (A x B x C x D x E) for all hands affected by the side count.
A = Frequency of player hand & dealer up card. Matchups that happen more often are more important. A good way to do this would be to use the frequency that a hand matchup occurs when tracking the count using the optimized modified tags for the play, from a deck composition representative of the true count at the index for the play. Doing the above was more work than I was willing to put into this, so I simply used the frequency of hands from an infinite deck. This isn’t ideal, but it’s far better than not including this factor at all.
B = Frequency of time the card distribution will meet or exceed the index for the play. These formulas are in the Theory of Blackjack. For example, if a play tends to happen at TC +3, then you’d use the frequency of time that the deck is at TC = 3 and all higher TCs. The closer to TC = 0 that an index is, the more impact that playing decision has. I used a composite frequency based on deck depth, utilizing the TC frequency distribution at 1/6, 2/6, 3/6, 4/6 and 5/6 into a 6D shoe.
C = Improvement in PE percentage using the modified tags, compared to the PE from the main system tags. Because a bigger PE improvement is better than a smaller PE improvement. Duh.
D = Rate of change of EV relative to TC near the index. As you meet and exceed the TC for an index play, some plays have EV that only changes slowly with the TC, and some plays the EV changes rapidly with TC. The more rapidly the EV changes with TC, the more impact that playing decision has.
E = Weighting for bet ramp. A playing decision that tends to happen with more money on the table has more impact. The weighting here can also be used to bifurcate the results into play-all and Wonging out (e.g. give a weight of zero for index plays you won’t be there for).
I lumped the insurance decision in with all the other plays.
So yeah, I did the math. Every possible balanced side count of 2 ranks was investigated and the magnitude of impact determined. I may not know how to write programs, but my skills with Excel and using the MGP CA are badass.
The best side count that came out of this analysis had an impact on ~150 of the top ~180 index plays, so this side count affects almost all the plays. Only two of the I18 indexes are not affected. The side count ranking was the same whether Wonging out or not. I then developed a full strategy for all indexes for this side count. I binned the affected plays into 4 different tag multipliers, making the overall strategy like a system of 5 systems. One thing is that when you combine the main tags with the side count tags, it is like using a different card counting system, so the index used when only using the main count is technically different than when using the modified tags. However, by lucky accident, nearly all the modified indexes were so similar that I can use the same indexes whether I am only using the main count, or the main count with side count, except for 2 very obscure plays where I’ve memorized 2 quite different indexes. There were no plays that used the side count alone. Memorizing which of the 4 possible tag adjustments go with each play was easy, because they make intuitive sense about deck composition and generally large sections of the strategy chart have the same tag multiplier. Opportunities to use the side count info to make better playing decisions happens with decent frequency, even in 8 deck, even on the first hand dealt (e.g. running count plays). The side count also improves the play of common side bets. I don’t have a proper sim of how this strategy performs, but I will try to work on that with a collaborator who can do this properly. It is my prediction that this system would probably be the strongest possible 2 parameter system that exists.
Bookmarks